The purpose of this work is to develop open-source software embedded in an affordable Raspberry Pi System, which can be installed in existing old vehicles and with the use of computer vision and other sensors have the ability to recognize and warn of impending traffic accidents or other dangerous situations, with the goal of the driver taking action and preventing them from escalating.

• Lane Departure Warning – (LDW)
  

• Forward Collision Warning – (FCW)
  

• Forward Vehicle Start Alarm – (FVSA)
  

• Red Light Violation Warning – (RLVW)
  

This program takes a feed from the camera or videofile and does Lane Departure Warning using houghlines and Forward Collision Warning, Forward Vehicle Start Alarm and Red Light Violation Warning using object detection (bonding boxes from mobile_net_v3_small with TensorFlow Lite)

Details of Software and Neural Network Model for Object Detection:

• Language: Python
• Framework: TensorFlow Lite
• Network: SSD MobileNet-V3_small
• Training Dataset: COCO Dataset

• Raspberry Pi 4 2gb or grater
• Usb or Pi Camera (logitech c920 recomended)
• Car Speed recognition (OBDII GPS Module can also be used (code not included))
• Turn signal recognition (Hall effect sensor)
• Status Lights (Blinkt! used)
• Power Supply (Usb Car Charger 15W+ recomended)
• Warning Buzzer (active buzzer recomended for clearer sound. Output jack 3.5 can also be used)

• Python 3.10 (bugs in 3.11+)
• opencv_python 4.6 or newer
• tensorflow 2.11 (not tested in later versions)
• pygame 2.1.2 or newer
• obd 0.7.1 or newer
• pyserial 3.5 or newer

python -m pip install -r requirements.txt 

Turn signal recognition (Hall effect sensor): Power 5V: Pin 2, Ground: Pin 20, Data: Pin 22 (GPIO 25)

Status Lights (Blinkt!): Power 5V: Pin 4, Ground: Pin 6, Data: Pins 16,18 (GPIO 23,24)

Warning Buzzer: Ground: Pin 39, Data: Pin 33 (GPIO 13)

Data pins can be adjusted inside the Pi-ADAS.py file

Via SSH Shell from Phone:

Bluetoothctl 
scan on 
pair XX:XX:XX:XX:XX:XX (mac address του OBD2) 
trust XX:XX:XX:XX:XX:XX
exit

xterm -hold -e 'sudo rfcomm bind rfcomm9 XX:XX:XX:XX:XX:XX && 
cd /home/pi/Desktop/Py-ADAS/ && 
/usr/bin/python /home/pi/Desktop/Py-ADAS/PI-ADAS.py'

Where XX:XX:XX:XX:XX:XX is mac address of OBDII

The Pi-ADAS.py file includes at the start all the debug states and configurable parameters of the system

Advanced driver-assistance system using Raspberry Pi, Python and OpenCV